Bringing Design to the Privacy Table: Broadening “Design” in “Privacy by Design” Through HCI [Paper Talk]
This post is based on a talk given at the 2019 ACM CHI Conference on Human Factors in Computing Systems (CHI 2019), in Glasgow, UK. The full research paper by Richmond Wong and Deirdre Mulligan that the talk is based on, “Bringing Design to the Privacy Table: Broadening “Design” in “Privacy by Design” Through the Lens of HCI” can be found here:
[Official ACM Version] [Open Access Pre-Print Version]
In our paper “Bringing Design to the Privacy Table: Broadening Design in Privacy by Design,” we conduct a curated literature review to make two conceptual argument arguments:
- There is a broad range of design practices used in human computer interaction (HCI) research which have been underutilized in Privacy By Design efforts.
- Broadening privacy by design’s notion of what “design” can do can help us more fully address privacy, particularly in situations where we don’t yet know what concepts or definitions of privacy are at stake.
But let me start with some background and motivation. I’m both a privacy researcher — studying studying how to develop technologies that respect privacy — and I’m a design researcher, who designs things to learn about the world.
I was excited several years ago to hear about a growing movement called “ Privacy By Design,” the idea that privacy protections should be embedded into products and organizational practice during the design of products, rather than trying to address privacy retroactively. Privacy By Design has been put forward in regulatory guidance from the US and other countries, and more recently by the EU’s General Data Protection Regulation. Yet these regulations don’t provide a lot of guidance about what Privacy By Design means in practice.
In interactions with and field observations of the interdisciplinary Privacy By Design community — including lawyers, regulators, academics, practitioners, and technical folks — I’ve found that there is a lot of recognition of the complexity of privacy: that it’s an essentially contested concept, there are many conceptualizations of privacy; privacy from companies is different than privacy from governments; there are different privacy harms, and so forth.
But the discussion of “design” seems much less complex. I had assumed Privacy By Design would have meant applying HCI’s rich breadth of design approaches toward privacy initiatives: user centered design, participatory design, value sensitive design, speculative design, and so on.
Instead, design seemed to be used narrowly, as either a way to implement the law via compliance engineering, or to solve specific privacy problems. Design was largely framed as a deductive way to solve a problem, using approaches such as encryption techniques or building systems to comply with fair information practices. While these are all important and necessary privacy initiatives, but I kept finding myself asking, “where’s the rest of design? “ Not just the deductive problem solving aspects of design, but also its the inductive, exploratory, and forward looking aspects.
There’s a gap here between the way the Privacy By Design views design and the way the HCI community views design. Since HCI researchers and practitioners are in a position to help support or implement privacy by design initiatives, it’s important to try to help broaden the notion of design in Privacy By Design to more fully bridge this gap.
So our paper aims to fulfill 2 goals:
- Design in HCI is more than just solving problems. We as HCI privacy researchers can more broadly engage the breadth of design approaches in HCI writ large. And there are opportunities to build connections among the HCI privacy research community and HCI design research community & research through design community to use design in relation to privacy in multiple ways.
- Privacy By Design efforts risk missing out on the full benefits that design can offer if it sticks with a narrower solution and compliance orientation to design. From HCI, we can help build bridges with interdisciplinary Privacy By Design community, and engage them in understanding a broader view of design.
So how might we characterize the breadth of ways that HCI uses design in relation to privacy? In the paper, we conduct a curated review of HCI research to explore and breadth and richness of how design practices are used in relation to privacy. We searched for HCI papers that use both the terms “privacy” and “design,” curating a corpus of 64 papers. Reading through each paper, we openly coded each one by asking a set of questions including: Why is design used; who is design done by; and for whom is design done? Using affinity diagramming on the open codes, we came up with a set of categories, or dimensions, which we used to re-code the corpus. In this post I’m going to focus on the dimensions that emerged when we looked at the “why design?” question, which we call the purposes of design.
We describe 4 purposes of design in relation to privacy. They are:
- Design to solve a privacy problem;
- Design to inform or support privacy;
- Design to explore people and situations; and
- Design to critique, speculate, and present critical alternatives.
Note that we use these to talk about how design has been used in privacy research specifically, not about all design writ large (that would be quite a different and broader endeavor!). In practice these categories are not mutually exclusive, and are not the only way to look at the space, but looking at them separately helps give some analytical clarity. Let’s briefly walk through each of these design purposes.
1. To Solve a Privacy Problem
First, design is seen as a way to solve a privacy problem — which occurred most often in the papers we looked at. And I think this is often how we think about design colloquially, as a set of practices to solve problems. This is often how design is discussed in Privacy By Design discussions as well.
When viewing design in this way, privacy is presented a problem that has already been well-defined at the before the design process, and a solution is designed to address that definition of the problem. A lot of responsibility for protecting privacy here is thus placed in the technical system.
For instance, if a problem of privacy is defined as the harms that result from long term data processing and aggregation, we might design a system that limits data retention. If a problem of privacy is defined as not being identified, we might design a system to be anonymous.
2. To Inform or Support Privacy
Second, design is seen as a way to inform or support actors who must make privacy-relevant choices, rather than solving a privacy problem outright. This was also common in our set of papers. Design to inform or support privacy views problems posed by privacy as an information or tools problem. If users receive information in better ways, or have better tools, then they can make more informed choices about how to act in privacy-preserving ways.
A lot of research has been done on how to design usable privacy policies or privacy notices — but it’s still up to the user to read the notice and make a privacy relevant decision. Other types of design work in this vein includes designing privacy icons, controls, dashboards, visualizations, as well as educational materials and activities.
In these approaches, a lot of responsibility for protecting privacy is placed in the choices that people make, informed by a design artifact. The protection of privacy doesn’t arise from the design of the system itself, but rather by how a person chooses to use the system. This orientation towards privacy fits well with US regulations around privacy that make individuals manage and control their own data.
3. To Explore People and Situations (Related to Privacy)
Third is using design to explore people and situations. Design is used as a mode of inquiry, to better understand what privacy or the experience of privacy means to certain people, in certain situations. Design here is not necessarily about solving an immediate problem.
Techniques like design probes or collaborative design workshops are some approaches here. For example, a project I presented at CSCW 2018 involved presenting booklets with conceptual designs of potentially invasive products to technology practitioners in training. We weren’t looking to gather feedback in order to develop these conceptual ideas into usable products. Instead, the goal was to use these conceptual design ideas as provocations to better understand the participants’ worldviews. How are they conceptualizing privacy when they see these designs? How do their reactions help us understand where they place responsibility for addressing privacy?
Here, privacy is understood as a situated experience, which emerges from practices from particular groups in specific contexts or situations. The goal is less about solving a privacy problem, and more about understanding how privacy gets enacted and experienced.
4. To Critique, Speculate, or Present Critical Alternatives About Privacy
Fourth is design to critique, speculate, or present critical alternatives. (By critical I don’t mean bad or mean, but instead I mean critical like reflexive reflection or careful analysis). Design here is not about exploring the world as it is, but focuses on how the world could be. Often this consists of creating create conceptual designs that provoke, to create a space to surface and discuss social values. These help us discuss worlds we might strive to achieve or ones we want to avoid. Privacy in this case is situated in different possible sociotechnical configurations of the world, thinking about privacy’s social, legal, and technical aspects together.
For example, in a project I presented at DIS 2017, we created advertisements for fictional sensing products, like a bodily implant for workplace employees. This helped us raise questions beyond basic data collection and use ones. The designs helped us ask questions about how is privacy implicated in the workplace, or through employment law? Can consent really occur with these power dynamics? It also helped us ask normative questions, such as: Who gets to have privacy and who doesn’t? Who or what should be responsible for protecting privacy? Might we look to technical design, to regulations, to market mechanisms, or to individual choice to protect privacy?
Design Is a Political, Values-Laden Choice
So in summary these are the 4 purposes of design that we identified in this paper: using design to solve, to inform and support, to explore, and to critique and speculate. Again, in practice, they’re not discrete categories. Many design approaches, like user centered design, or participatory design, use design for multiple design purposes.
But this variety of purposes for how design relates to privacy is also a reminder that design isn’t a neutral process, but is itself political and values-laden. (Not political in terms of liberal and conservative, but political in the sense that there is power and social implications in the choices we make about how to use design). Each design purpose suggests a different starting place for how we orient ourselves towards conceptualizing and operationalizing privacy. We might think about privacy as:
- a technical property;
- a user-made choice;
- situated experiences;
- sociotechnically situated.
Privacy can be many and all of these things at once, but the design methods we choose, and the reasons why we choose to use design helps to suggest or foreclose different orientations toward privacy. These choices also suggest that responsibility for privacy might be placed in different places — such as in a technical system, in a person’s choices, in a platform’s policies, in the law, in the market, and so forth.
Now I’ve been discussing these 4 design purposes equally, but they weren’t equal in our corpus. Allowing each paper to be coded for multiple categories, a little over half the papers we looked at used design to solve a privacy problem and a little over half used design to inform or support. Less than a quarter used design to explore; even fewer used design to critique and speculate. We don’t claim that the exact percentages are representative of all the privacy literature, but there’s a qualitative difference here, where most of the work we reviewed uses design to solve privacy problems or support and inform privacy.
We are arguing for a big tent approach in privacy by design: using design in all of these ways helps us address a broader set of conceptions of privacy.
This suggests that there’s an opportunity for us to build bridges between the HCI privacy research community, which has rich domain expertise; and the HCI design research & research through design communities, which have rich design methods expertise, particularly using design in ways to explore, and to critique and speculate.
So that’s Argument 1, that we have the opportunity to build new bridges among HCI communities to more fully make use of each others’ expertise, and a broader range of design methods and purposes.
Argument 2 is that Privacy By Design has largely (with some exceptions) thought about design as a problem solving process. Privacy By Design research and practice could expand on that thinking of design to make more use of a fuller breadth of uses of design that are reflected in HCI.
Implications for Design Collaboration
So what might some of these collaborations within and across fields look like, if we want to make use of more of design’s breadth? For example if we as privacy researchers, develop a set of usable privacy tools to inform and support most people’s privacy decision making; that might be complemented with design to explore so that we can better understand the often marginalized populations for whom those tools don’t work. For instance Diana Freed et al.’s work shows that social media privacy and security tools can be used against victims of intimate partner violence, violating their privacy and safety. Or, an emerging set of problems we face is thinking about privacy in physically instrumented spaces: how does consent work, what conceptions of privacy and privacy risk are at play? We can complement design to solve and design to support efforts with design to critique and speculate; to craft future scenarios that try to understand what concepts of privacy might be at play, and how privacy can surface differently when technical, social, or legal aspects of the world change.
From a design research perspective, I think there’s growing interest in the design research community to create provocative artifacts to try to surface discussions about privacy, particularly in relation to new and emerging technologies. Critically reflecting on my own design research work, I think it can be tempting to just speak to other designers and resort to conceptions of privacy that say “surveillance is creepy” and not dig deeper into other approaches to privacy. But by collaborating with privacy researchers, we can bring more domain expertise and theoretical depth to these design explorations and speculations, and engage a broader set of privacy stakeholders.
Industry privacy practitioners working on privacy by design initiatives might consider incorporating more UX researchers and designers form their organizations, as privacy allies and as design experts. Approaches that use design to critique and speculate may also align well with privacy practitioners’ stated desire to find contextual and anticipatory privacy tools to help “ think around corners “, as reported by Ken Bamberger and Deirdre Mulligan.
Privacy By Design regulators could incorporate more designers (in addition to engineers and computer scientists) in regulatory discussions about privacy by design, so that this richness of design practice isn’t lost when the words “by design” are written in the law.
Moreover, there’s an opportunity here for us an HCI community to bring HCI’s rich notions of what design can mean to Privacy By Design, so that beyond being a problem solving process, it is also seen as a process that also makes use of the multi-faceted, inductive, and exploratory uses of design that this community engages in.
Paper Citation: Richmond Y. Wong and Deirdre K. Mulligan. 2019. Bringing Design to the Privacy Table: Broadening “Design” in “Privacy by Design” Through the Lens of HCI. In Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems (CHI ‘19). ACM, New York, NY, USA, Paper 262, 17 pages. DOI: https://doi.org/10.1145/3290605.3300492
Originally published at http://bytegeist.wordpress.com on May 10, 2019.